Add experimental wasm bindings

Published to npm as objdiff-wasm

objdiff-core/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "objdiff-core"
-version = "2.0.0-beta.4"
+version = "2.0.0-beta.5"
 edition = "2021"
 rust-version = "1.70"
 authors = ["Luke Street <luke@street.dev>"]
@@ -11,6 +11,9 @@ description = """
 A local diffing tool for decompilation projects.
 """
 
+[lib]
+crate-type = ["cdylib", "rlib"]
+
 [features]
 all = ["config", "dwarf", "mips", "ppc", "x86", "arm"]
 any-arch = [] # Implicit, used to check if any arch is enabled
@@ -20,6 +23,7 @@ mips = ["any-arch", "rabbitizer"]
 ppc = ["any-arch", "cwdemangle", "cwextab", "ppc750cl"]
 x86 = ["any-arch", "cpp_demangle", "iced-x86", "msvc-demangler"]
 arm = ["any-arch", "cpp_demangle", "unarm", "arm-attr"]
+wasm = ["serde_json"]
 
 [dependencies]
 anyhow = "1.0.82"
@@ -30,9 +34,12 @@ log = "0.4.21"
 memmap2 = "0.9.4"
 num-traits = "0.2.18"
 object = { version = "0.35.0", features = ["read_core", "std", "elf", "pe"], default-features = false }
+pbjson = "0.7.0"
+prost = "0.13.1"
 serde = { version = "1", features = ["derive"] }
 similar = { version = "2.5.0", default-features = false }
 strum = { version = "0.26.2", features = ["derive"] }
+wasm-bindgen = "0.2.93"
 
 # config
 globset = { version = "0.4.14", features = ["serde1"], optional = true }
@@ -59,3 +66,7 @@ msvc-demangler = { version = "0.10.0", optional = true }
 # arm
 unarm = { version = "1.4.0", optional = true }
 arm-attr = { version = "0.1.1", optional = true }
+
+[build-dependencies]
+prost-build = "0.13.1"
+pbjson-build = "0.7.0"

objdiff-core/build.rs

@@ -0,0 +1,54 @@
+use std::path::{Path, PathBuf};
+
+fn main() {
+    let root = PathBuf::from(env!("CARGO_MANIFEST_DIR")).join("protos");
+    let descriptor_path = root.join("proto_descriptor.bin");
+    println!("cargo:rerun-if-changed={}", descriptor_path.display());
+    let descriptor_mtime = std::fs::metadata(&descriptor_path)
+        .map(|m| m.modified().unwrap())
+        .unwrap_or(std::time::SystemTime::UNIX_EPOCH);
+    let mut run_protoc = false;
+    let proto_files = vec![root.join("diff.proto"), root.join("report.proto")];
+    for proto_file in &proto_files {
+        println!("cargo:rerun-if-changed={}", proto_file.display());
+        let mtime = match std::fs::metadata(proto_file) {
+            Ok(m) => m.modified().unwrap(),
+            Err(e) => panic!("Failed to stat proto file {}: {:?}", proto_file.display(), e),
+        };
+        if mtime > descriptor_mtime {
+            run_protoc = true;
+        }
+    }
+
+    fn prost_config(descriptor_path: &Path, run_protoc: bool) -> prost_build::Config {
+        let mut config = prost_build::Config::new();
+        config.file_descriptor_set_path(descriptor_path);
+        // If our cached descriptor is up-to-date, we don't need to run protoc.
+        // This is helpful so that users don't need to have protoc installed
+        // unless they're updating the protos.
+        if !run_protoc {
+            config.skip_protoc_run();
+        }
+        config
+    }
+    if let Err(e) =
+        prost_config(&descriptor_path, run_protoc).compile_protos(&proto_files, &[root.as_path()])
+    {
+        if e.kind() == std::io::ErrorKind::NotFound && e.to_string().contains("protoc") {
+            eprintln!("protoc not found, skipping protobuf compilation");
+            prost_config(&descriptor_path, false)
+                .compile_protos(&proto_files, &[root.as_path()])
+                .expect("Failed to compile protos");
+        } else {
+            panic!("Failed to compile protos: {e:?}");
+        }
+    }
+
+    let descriptor_set = std::fs::read(descriptor_path).expect("Failed to read descriptor set");
+    pbjson_build::Builder::new()
+        .register_descriptors(&descriptor_set)
+        .expect("Failed to register descriptors")
+        .preserve_proto_field_names()
+        .build(&[".objdiff"])
+        .expect("Failed to build pbjson");
+}

objdiff-core/protos/diff.proto

@@ -0,0 +1,163 @@
+syntax = "proto3";
+
+package objdiff.diff;
+
+// A symbol
+message Symbol {
+  // Name of the symbol
+  string name = 1;
+  // Demangled name of the symbol
+  optional string demangled_name = 2;
+  // Symbol address
+  uint64 address = 3;
+  // Symbol size
+  uint64 size = 4;
+  // Bitmask of SymbolFlag
+  uint32 flags = 5;
+}
+
+// Symbol visibility flags
+enum SymbolFlag {
+  SYMBOL_NONE = 0;
+  SYMBOL_GLOBAL = 1;
+  SYMBOL_LOCAL = 2;
+  SYMBOL_WEAK = 3;
+  SYMBOL_COMMON = 4;
+  SYMBOL_HIDDEN = 5;
+}
+
+// A single parsed instruction
+message Instruction {
+  // Instruction address
+  uint64 address = 1;
+  // Instruction size
+  uint32 size = 2;
+  // Instruction opcode
+  uint32 opcode = 3;
+  // Instruction mnemonic
+  string mnemonic = 4;
+  // Instruction formatted string
+  string formatted = 5;
+  // Original (unsimplified) instruction string
+  optional string original = 6;
+  // Instruction arguments
+  repeated Argument arguments = 7;
+  // Instruction relocation
+  optional Relocation relocation = 8;
+  // Instruction branch destination
+  optional uint64 branch_dest = 9;
+  // Instruction line number
+  optional uint32 line_number = 10;
+}
+
+// An instruction argument
+message Argument {
+  oneof value {
+    // Plain text
+    string plain_text = 1;
+    // Value
+    ArgumentValue argument = 2;
+    // Relocation
+    ArgumentRelocation relocation = 3;
+    // Branch destination
+    uint64 branch_dest = 4;
+  }
+}
+
+// An instruction argument value
+message ArgumentValue {
+  oneof value {
+    // Signed integer
+    int64 signed = 1;
+    // Unsigned integer
+    uint64 unsigned = 2;
+    // Opaque value
+    string opaque = 3;
+  }
+}
+
+// Marker type for relocation arguments
+message ArgumentRelocation {
+}
+
+message Relocation {
+  uint32 type = 1;
+  string type_name = 2;
+  RelocationTarget target = 3;
+}
+
+message RelocationTarget {
+  Symbol symbol = 1;
+  int64 addend = 2;
+}
+
+message InstructionDiff {
+  DiffKind diff_kind = 1;
+  optional Instruction instruction = 2;
+  optional InstructionBranchFrom branch_from = 3;
+  optional InstructionBranchTo branch_to = 4;
+  repeated ArgumentDiff arg_diff = 5;
+}
+
+message ArgumentDiff {
+  optional uint32 diff_index = 1;
+}
+
+enum DiffKind {
+  DIFF_NONE = 0;
+  DIFF_REPLACE = 1;
+  DIFF_DELETE = 2;
+  DIFF_INSERT = 3;
+  DIFF_OP_MISMATCH = 4;
+  DIFF_ARG_MISMATCH = 5;
+}
+
+message InstructionBranchFrom {
+  repeated uint32 instruction_index = 1;
+  uint32 branch_index = 2;
+}
+
+message InstructionBranchTo {
+  uint32 instruction_index = 1;
+  uint32 branch_index = 2;
+}
+
+message FunctionDiff {
+  Symbol symbol = 1;
+  repeated InstructionDiff instructions = 2;
+  optional float match_percent = 3;
+}
+
+message DataDiff {
+  DiffKind kind = 1;
+  bytes data = 2;
+  // May be larger than data
+  uint64 size = 3;
+}
+
+message SectionDiff {
+  string name = 1;
+  SectionKind kind = 2;
+  uint64 size = 3;
+  uint64 address = 4;
+  repeated FunctionDiff functions = 5;
+  repeated DataDiff data = 6;
+  optional float match_percent = 7;
+}
+
+enum SectionKind {
+  SECTION_UNKNOWN = 0;
+  SECTION_TEXT = 1;
+  SECTION_DATA = 2;
+  SECTION_BSS = 3;
+  SECTION_COMMON = 4;
+}
+
+message ObjectDiff {
+  repeated SectionDiff sections = 1;
+}
+
+message DiffResult {
+  optional ObjectDiff left = 1;
+  optional ObjectDiff right = 2;
+}

objdiff-core/protos/report.proto

@@ -0,0 +1,135 @@
+syntax = "proto3";
+
+package objdiff.report;
+
+// Progress info for a report or unit
+message Measures {
+  // Overall match percent, including partially matched functions and data
+  float fuzzy_match_percent = 1;
+  // Total size of code in bytes
+  uint64 total_code = 2;
+  // Fully matched code size in bytes
+  uint64 matched_code = 3;
+  // Fully matched code percent
+  float matched_code_percent = 4;
+  // Total size of data in bytes
+  uint64 total_data = 5;
+  // Fully matched data size in bytes
+  uint64 matched_data = 6;
+  // Fully matched data percent
+  float matched_data_percent = 7;
+  // Total number of functions
+  uint32 total_functions = 8;
+  // Fully matched functions
+  uint32 matched_functions = 9;
+  // Fully matched functions percent
+  float matched_functions_percent = 10;
+}
+
+// Project progress report
+message Report {
+  // Overall progress info
+  Measures measures = 1;
+  // Units within this report
+  repeated ReportUnit units = 2;
+}
+
+// A unit of the report (usually a translation unit)
+message ReportUnit {
+  // The name of the unit
+  string name = 1;
+  // Progress info for this unit
+  Measures measures = 2;
+  // Sections within this unit
+  repeated ReportItem sections = 3;
+  // Functions within this unit
+  repeated ReportItem functions = 4;
+  // Extra metadata for this unit
+  optional ReportUnitMetadata metadata = 5;
+}
+
+// Extra metadata for a unit
+message ReportUnitMetadata {
+  // Whether this unit is marked as complete (or "linked")
+  optional bool complete = 1;
+  // The name of the module this unit belongs to
+  optional string module_name = 2;
+  // The ID of the module this unit belongs to
+  optional uint32 module_id = 3;
+  // The path to the source file of this unit
+  optional string source_path = 4;
+}
+
+// A section or function within a unit
+message ReportItem {
+  // The name of the item
+  string name = 1;
+  // The size of the item in bytes
+  uint64 size = 2;
+  // The overall match percent for this item
+  float fuzzy_match_percent = 3;
+  // Extra metadata for this item
+  optional ReportItemMetadata metadata = 4;
+}
+
+// Extra metadata for an item
+message ReportItemMetadata {
+  // The demangled name of the function
+  optional string demangled_name = 1;
+  // The virtual address of the function or section
+  optional uint64 virtual_address = 2;
+}
+
+// A pair of reports to compare and generate changes
+message ChangesInput {
+  // The previous report
+  Report from = 1;
+  // The current report
+  Report to = 2;
+}
+
+// Changes between two reports
+message Changes {
+  // The progress info for the previous report
+  Measures from = 1;
+  // The progress info for the current report
+  Measures to = 2;
+  // Units that changed
+  repeated ChangeUnit units = 3;
+}
+
+// A changed unit
+message ChangeUnit {
+  // The name of the unit
+  string name = 1;
+  // The previous progress info (omitted if new)
+  optional Measures from = 2;
+  // The current progress info (omitted if removed)
+  optional Measures to = 3;
+  // Sections that changed
+  repeated ChangeItem sections = 4;
+  // Functions that changed
+  repeated ChangeItem functions = 5;
+  // Extra metadata for this unit
+  optional ReportUnitMetadata metadata = 6;
+}
+
+// A changed section or function
+message ChangeItem {
+  // The name of the item
+  string name = 1;
+  // The previous progress info (omitted if new)
+  optional ChangeItemInfo from = 2;
+  // The current progress info (omitted if removed)
+  optional ChangeItemInfo to = 3;
+  // Extra metadata for this item
+  optional ReportItemMetadata metadata = 4;
+}
+
+// Progress info for a section or function
+message ChangeItemInfo {
+  // The overall match percent for this item
+  float fuzzy_match_percent = 1;
+  // The size of the item in bytes
+  uint64 size = 2;
+}

objdiff-core/src/arch/arm.rs

@@ -111,7 +111,7 @@ impl ObjArch for ObjArchArm {
         code: &[u8],
         section_index: usize,
         relocations: &[ObjReloc],
-        line_info: &BTreeMap<u64, u64>,
+        line_info: &BTreeMap<u64, u32>,
         config: &DiffObjConfig,
     ) -> Result<ProcessCodeResult> {
         let start_addr = address as u32;

objdiff-core/src/arch/mips.rs

@@ -85,7 +85,7 @@ impl ObjArch for ObjArchMips {
         code: &[u8],
         _section_index: usize,
         relocations: &[ObjReloc],
-        line_info: &BTreeMap<u64, u64>,
+        line_info: &BTreeMap<u64, u32>,
         config: &DiffObjConfig,
     ) -> Result<ProcessCodeResult> {
         let _guard = RABBITIZER_MUTEX.lock().map_err(|e| anyhow!("Failed to lock mutex: {e}"))?;

objdiff-core/src/arch/mod.rs

@@ -24,7 +24,7 @@ pub trait ObjArch: Send + Sync {
         code: &[u8],
         section_index: usize,
         relocations: &[ObjReloc],
-        line_info: &BTreeMap<u64, u64>,
+        line_info: &BTreeMap<u64, u32>,
         config: &DiffObjConfig,
     ) -> Result<ProcessCodeResult>;
 

objdiff-core/src/arch/ppc.rs

@@ -35,7 +35,7 @@ impl ObjArch for ObjArchPpc {
         code: &[u8],
         _section_index: usize,
         relocations: &[ObjReloc],
-        line_info: &BTreeMap<u64, u64>,
+        line_info: &BTreeMap<u64, u32>,
         config: &DiffObjConfig,
     ) -> Result<ProcessCodeResult> {
         let ins_count = code.len() / 4;

objdiff-core/src/arch/x86.rs

@@ -32,7 +32,7 @@ impl ObjArch for ObjArchX86 {
         code: &[u8],
         _section_index: usize,
         relocations: &[ObjReloc],
-        line_info: &BTreeMap<u64, u64>,
+        line_info: &BTreeMap<u64, u32>,
         config: &DiffObjConfig,
     ) -> Result<ProcessCodeResult> {
         let mut result = ProcessCodeResult { ops: Vec::new(), insts: Vec::new() };

objdiff-core/src/bindings/diff.rs

@@ -0,0 +1,244 @@
+use crate::{
+    diff::{
+        ObjDataDiff, ObjDataDiffKind, ObjDiff, ObjInsArgDiff, ObjInsBranchFrom, ObjInsBranchTo,
+        ObjInsDiff, ObjInsDiffKind, ObjSectionDiff, ObjSymbolDiff,
+    },
+    obj::{
+        ObjInfo, ObjIns, ObjInsArg, ObjInsArgValue, ObjReloc, ObjSectionKind, ObjSymbol,
+        ObjSymbolFlagSet, ObjSymbolFlags,
+    },
+};
+
+// Protobuf diff types
+include!(concat!(env!("OUT_DIR"), "/objdiff.diff.rs"));
+include!(concat!(env!("OUT_DIR"), "/objdiff.diff.serde.rs"));
+
+impl DiffResult {
+    pub fn new(left: Option<(&ObjInfo, &ObjDiff)>, right: Option<(&ObjInfo, &ObjDiff)>) -> Self {
+        Self {
+            left: left.map(|(obj, diff)| ObjectDiff::new(obj, diff)),
+            right: right.map(|(obj, diff)| ObjectDiff::new(obj, diff)),
+        }
+    }
+}
+
+impl ObjectDiff {
+    pub fn new(obj: &ObjInfo, diff: &ObjDiff) -> Self {
+        Self {
+            sections: diff
+                .sections
+                .iter()
+                .enumerate()
+                .map(|(i, d)| SectionDiff::new(obj, i, d))
+                .collect(),
+        }
+    }
+}
+
+impl SectionDiff {
+    pub fn new(obj: &ObjInfo, section_index: usize, section_diff: &ObjSectionDiff) -> Self {
+        let section = &obj.sections[section_index];
+        let functions = section_diff.symbols.iter().map(|d| FunctionDiff::new(obj, d)).collect();
+        let data = section_diff.data_diff.iter().map(|d| DataDiff::new(obj, d)).collect();
+        Self {
+            name: section.name.to_string(),
+            kind: SectionKind::from(section.kind) as i32,
+            size: section.size,
+            address: section.address,
+            functions,
+            data,
+            match_percent: section_diff.match_percent,
+        }
+    }
+}
+
+impl From<ObjSectionKind> for SectionKind {
+    fn from(value: ObjSectionKind) -> Self {
+        match value {
+            ObjSectionKind::Code => SectionKind::SectionText,
+            ObjSectionKind::Data => SectionKind::SectionData,
+            ObjSectionKind::Bss => SectionKind::SectionBss,
+            // TODO common
+        }
+    }
+}
+
+impl FunctionDiff {
+    pub fn new(object: &ObjInfo, symbol_diff: &ObjSymbolDiff) -> Self {
+        let (_section, symbol) = object.section_symbol(symbol_diff.symbol_ref);
+        // let diff_symbol = symbol_diff.diff_symbol.map(|symbol_ref| {
+        //     let (_section, symbol) = object.section_symbol(symbol_ref);
+        //     Symbol::from(symbol)
+        // });
+        let instructions = symbol_diff.instructions.iter().map(InstructionDiff::from).collect();
+        Self {
+            symbol: Some(Symbol::from(symbol)),
+            // diff_symbol,
+            instructions,
+            match_percent: symbol_diff.match_percent,
+        }
+    }
+}
+
+impl DataDiff {
+    pub fn new(_object: &ObjInfo, data_diff: &ObjDataDiff) -> Self {
+        Self {
+            kind: DiffKind::from(data_diff.kind) as i32,
+            data: data_diff.data.clone(),
+            size: data_diff.len as u64,
+        }
+    }
+}
+
+impl<'a> From<&'a ObjSymbol> for Symbol {
+    fn from(value: &'a ObjSymbol) -> Self {
+        Self {
+            name: value.name.to_string(),
+            demangled_name: value.demangled_name.clone(),
+            address: value.address,
+            size: value.size,
+            flags: symbol_flags(value.flags),
+        }
+    }
+}
+
+fn symbol_flags(value: ObjSymbolFlagSet) -> u32 {
+    let mut flags = 0u32;
+    if value.0.contains(ObjSymbolFlags::Global) {
+        flags |= SymbolFlag::SymbolNone as u32;
+    }
+    if value.0.contains(ObjSymbolFlags::Local) {
+        flags |= SymbolFlag::SymbolLocal as u32;
+    }
+    if value.0.contains(ObjSymbolFlags::Weak) {
+        flags |= SymbolFlag::SymbolWeak as u32;
+    }
+    if value.0.contains(ObjSymbolFlags::Common) {
+        flags |= SymbolFlag::SymbolCommon as u32;
+    }
+    if value.0.contains(ObjSymbolFlags::Hidden) {
+        flags |= SymbolFlag::SymbolHidden as u32;
+    }
+    flags
+}
+
+impl<'a> From<&'a ObjIns> for Instruction {
+    fn from(value: &'a ObjIns) -> Self {
+        Self {
+            address: value.address,
+            size: value.size as u32,
+            opcode: value.op as u32,
+            mnemonic: value.mnemonic.clone(),
+            formatted: value.formatted.clone(),
+            arguments: value.args.iter().map(Argument::from).collect(),
+            relocation: value.reloc.as_ref().map(Relocation::from),
+            branch_dest: value.branch_dest,
+            line_number: value.line,
+            original: value.orig.clone(),
+        }
+    }
+}
+
+impl<'a> From<&'a ObjInsArg> for Argument {
+    fn from(value: &'a ObjInsArg) -> Self {
+        Self {
+            value: Some(match value {
+                ObjInsArg::PlainText(s) => argument::Value::PlainText(s.to_string()),
+                ObjInsArg::Arg(v) => argument::Value::Argument(ArgumentValue::from(v)),
+                ObjInsArg::Reloc => argument::Value::Relocation(ArgumentRelocation {}),
+                ObjInsArg::BranchDest(dest) => argument::Value::BranchDest(*dest),
+            }),
+        }
+    }
+}
+
+impl From<&ObjInsArgValue> for ArgumentValue {
+    fn from(value: &ObjInsArgValue) -> Self {
+        Self {
+            value: Some(match value {
+                ObjInsArgValue::Signed(v) => argument_value::Value::Signed(*v),
+                ObjInsArgValue::Unsigned(v) => argument_value::Value::Unsigned(*v),
+                ObjInsArgValue::Opaque(v) => argument_value::Value::Opaque(v.to_string()),
+            }),
+        }
+    }
+}
+
+impl<'a> From<&'a ObjReloc> for Relocation {
+    fn from(value: &ObjReloc) -> Self {
+        Self {
+            r#type: match value.flags {
+                object::RelocationFlags::Elf { r_type } => r_type,
+                object::RelocationFlags::MachO { r_type, .. } => r_type as u32,
+                object::RelocationFlags::Coff { typ } => typ as u32,
+                object::RelocationFlags::Xcoff { r_rtype, .. } => r_rtype as u32,
+                _ => unreachable!(),
+            },
+            type_name: String::new(), // TODO
+            target: Some(RelocationTarget::from(&value.target)),
+        }
+    }
+}
+
+impl<'a> From<&'a ObjSymbol> for RelocationTarget {
+    fn from(value: &'a ObjSymbol) -> Self {
+        Self { symbol: Some(Symbol::from(value)), addend: value.addend }
+    }
+}
+
+impl<'a> From<&'a ObjInsDiff> for InstructionDiff {
+    fn from(value: &'a ObjInsDiff) -> Self {
+        Self {
+            instruction: value.ins.as_ref().map(Instruction::from),
+            diff_kind: DiffKind::from(value.kind) as i32,
+            branch_from: value.branch_from.as_ref().map(InstructionBranchFrom::from),
+            branch_to: value.branch_to.as_ref().map(InstructionBranchTo::from),
+            arg_diff: value.arg_diff.iter().map(ArgumentDiff::from).collect(),
+        }
+    }
+}
+
+impl From<&Option<ObjInsArgDiff>> for ArgumentDiff {
+    fn from(value: &Option<ObjInsArgDiff>) -> Self {
+        Self { diff_index: value.as_ref().map(|v| v.idx as u32) }
+    }
+}
+
+impl From<ObjInsDiffKind> for DiffKind {
+    fn from(value: ObjInsDiffKind) -> Self {
+        match value {
+            ObjInsDiffKind::None => DiffKind::DiffNone,
+            ObjInsDiffKind::OpMismatch => DiffKind::DiffOpMismatch,
+            ObjInsDiffKind::ArgMismatch => DiffKind::DiffArgMismatch,
+            ObjInsDiffKind::Replace => DiffKind::DiffReplace,
+            ObjInsDiffKind::Delete => DiffKind::DiffDelete,
+            ObjInsDiffKind::Insert => DiffKind::DiffInsert,
+        }
+    }
+}
+
+impl From<ObjDataDiffKind> for DiffKind {
+    fn from(value: ObjDataDiffKind) -> Self {
+        match value {
+            ObjDataDiffKind::None => DiffKind::DiffNone,
+            ObjDataDiffKind::Replace => DiffKind::DiffReplace,
+            ObjDataDiffKind::Delete => DiffKind::DiffDelete,
+            ObjDataDiffKind::Insert => DiffKind::DiffInsert,
+        }
+    }
+}
+
+impl<'a> From<&'a ObjInsBranchFrom> for InstructionBranchFrom {
+    fn from(value: &'a ObjInsBranchFrom) -> Self {
+        Self {
+            instruction_index: value.ins_idx.iter().map(|&x| x as u32).collect(),
+            branch_index: value.branch_idx as u32,
+        }
+    }
+}
+
+impl<'a> From<&'a ObjInsBranchTo> for InstructionBranchTo {
+    fn from(value: &'a ObjInsBranchTo) -> Self {
+        Self { instruction_index: value.ins_idx as u32, branch_index: value.branch_idx as u32 }
+    }
+}

objdiff-core/src/bindings/mod.rs

@@ -0,0 +1,4 @@
+pub mod diff;
+pub mod report;
+#[cfg(feature = "wasm")]
+pub mod wasm;

objdiff-core/src/bindings/report.rs

@@ -0,0 +1,232 @@
+use anyhow::{bail, Result};
+use prost::Message;
+use serde_json::error::Category;
+
+// Protobuf report types
+include!(concat!(env!("OUT_DIR"), "/objdiff.report.rs"));
+include!(concat!(env!("OUT_DIR"), "/objdiff.report.serde.rs"));
+
+impl Report {
+    pub fn parse(data: &[u8]) -> Result<Self> {
+        if data.is_empty() {
+            bail!(std::io::Error::from(std::io::ErrorKind::UnexpectedEof));
+        }
+        if data[0] == b'{' {
+            // Load as JSON
+            Self::from_json(data).map_err(anyhow::Error::new)
+        } else {
+            // Load as binary protobuf
+            Self::decode(data).map_err(anyhow::Error::new)
+        }
+    }
+
+    fn from_json(bytes: &[u8]) -> Result<Self, serde_json::Error> {
+        match serde_json::from_slice::<Self>(bytes) {
+            Ok(report) => Ok(report),
+            Err(e) => {
+                match e.classify() {
+                    Category::Io | Category::Eof | Category::Syntax => Err(e),
+                    Category::Data => {
+                        // Try to load as legacy report
+                        match serde_json::from_slice::<LegacyReport>(bytes) {
+                            Ok(legacy_report) => Ok(Report::from(legacy_report)),
+                            Err(_) => Err(e),
+                        }
+                    }
+                }
+            }
+        }
+    }
+}
+
+impl Measures {
+    /// Average the fuzzy match percentage over total code bytes.
+    pub fn calc_fuzzy_match_percent(&mut self) {
+        if self.total_code == 0 {
+            self.fuzzy_match_percent = 100.0;
+        } else {
+            self.fuzzy_match_percent /= self.total_code as f32;
+        }
+    }
+
+    /// Calculate the percentage of matched code, data, and functions.
+    pub fn calc_matched_percent(&mut self) {
+        self.matched_code_percent = if self.total_code == 0 {
+            100.0
+        } else {
+            self.matched_code as f32 / self.total_code as f32 * 100.0
+        };
+        self.matched_data_percent = if self.total_data == 0 {
+            100.0
+        } else {
+            self.matched_data as f32 / self.total_data as f32 * 100.0
+        };
+        self.matched_functions_percent = if self.total_functions == 0 {
+            100.0
+        } else {
+            self.matched_functions as f32 / self.total_functions as f32 * 100.0
+        };
+    }
+}
+
+impl From<&ReportItem> for ChangeItemInfo {
+    fn from(value: &ReportItem) -> Self {
+        Self { fuzzy_match_percent: value.fuzzy_match_percent, size: value.size }
+    }
+}
+
+/// Allows [collect](Iterator::collect) to be used on an iterator of [Measures].
+impl FromIterator<Measures> for Measures {
+    fn from_iter<T>(iter: T) -> Self
+    where T: IntoIterator<Item = Measures> {
+        let mut measures = Measures::default();
+        for other in iter {
+            measures.fuzzy_match_percent += other.fuzzy_match_percent * other.total_code as f32;
+            measures.total_code += other.total_code;
+            measures.matched_code += other.matched_code;
+            measures.total_data += other.total_data;
+            measures.matched_data += other.matched_data;
+            measures.total_functions += other.total_functions;
+            measures.matched_functions += other.matched_functions;
+        }
+        measures.calc_fuzzy_match_percent();
+        measures.calc_matched_percent();
+        measures
+    }
+}
+
+// Older JSON report types
+#[derive(Debug, Clone, Default, serde::Serialize, serde::Deserialize)]
+struct LegacyReport {
+    fuzzy_match_percent: f32,
+    total_code: u64,
+    matched_code: u64,
+    matched_code_percent: f32,
+    total_data: u64,
+    matched_data: u64,
+    matched_data_percent: f32,
+    total_functions: u32,
+    matched_functions: u32,
+    matched_functions_percent: f32,
+    units: Vec<LegacyReportUnit>,
+}
+
+impl From<LegacyReport> for Report {
+    fn from(value: LegacyReport) -> Self {
+        Self {
+            measures: Some(Measures {
+                fuzzy_match_percent: value.fuzzy_match_percent,
+                total_code: value.total_code,
+                matched_code: value.matched_code,
+                matched_code_percent: value.matched_code_percent,
+                total_data: value.total_data,
+                matched_data: value.matched_data,
+                matched_data_percent: value.matched_data_percent,
+                total_functions: value.total_functions,
+                matched_functions: value.matched_functions,
+                matched_functions_percent: value.matched_functions_percent,
+            }),
+            units: value.units.into_iter().map(ReportUnit::from).collect(),
+        }
+    }
+}
+
+#[derive(Debug, Clone, Default, serde::Serialize, serde::Deserialize)]
+struct LegacyReportUnit {
+    name: String,
+    fuzzy_match_percent: f32,
+    total_code: u64,
+    matched_code: u64,
+    total_data: u64,
+    matched_data: u64,
+    total_functions: u32,
+    matched_functions: u32,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    complete: Option<bool>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    module_name: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    module_id: Option<u32>,
+    sections: Vec<LegacyReportItem>,
+    functions: Vec<LegacyReportItem>,
+}
+
+impl From<LegacyReportUnit> for ReportUnit {
+    fn from(value: LegacyReportUnit) -> Self {
+        let mut measures = Measures {
+            fuzzy_match_percent: value.fuzzy_match_percent,
+            total_code: value.total_code,
+            matched_code: value.matched_code,
+            total_data: value.total_data,
+            matched_data: value.matched_data,
+            total_functions: value.total_functions,
+            matched_functions: value.matched_functions,
+            ..Default::default()
+        };
+        measures.calc_matched_percent();
+        Self {
+            name: value.name.clone(),
+            measures: Some(measures),
+            sections: value.sections.into_iter().map(ReportItem::from).collect(),
+            functions: value.functions.into_iter().map(ReportItem::from).collect(),
+            metadata: Some(ReportUnitMetadata {
+                complete: value.complete,
+                module_name: value.module_name.clone(),
+                module_id: value.module_id,
+                ..Default::default()
+            }),
+        }
+    }
+}
+
+#[derive(Debug, Clone, Default, serde::Serialize, serde::Deserialize)]
+struct LegacyReportItem {
+    name: String,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    demangled_name: Option<String>,
+    #[serde(
+        default,
+        skip_serializing_if = "Option::is_none",
+        serialize_with = "serialize_hex",
+        deserialize_with = "deserialize_hex"
+    )]
+    address: Option<u64>,
+    size: u64,
+    fuzzy_match_percent: f32,
+}
+
+impl From<LegacyReportItem> for ReportItem {
+    fn from(value: LegacyReportItem) -> Self {
+        Self {
+            name: value.name,
+            size: value.size,
+            fuzzy_match_percent: value.fuzzy_match_percent,
+            metadata: Some(ReportItemMetadata {
+                demangled_name: value.demangled_name,
+                virtual_address: value.address,
+            }),
+        }
+    }
+}
+
+fn serialize_hex<S>(x: &Option<u64>, s: S) -> Result<S::Ok, S::Error>
+where S: serde::Serializer {
+    if let Some(x) = x {
+        s.serialize_str(&format!("{:#x}", x))
+    } else {
+        s.serialize_none()
+    }
+}
+
+fn deserialize_hex<'de, D>(d: D) -> Result<Option<u64>, D::Error>
+where D: serde::Deserializer<'de> {
+    use serde::Deserialize;
+    let s = String::deserialize(d)?;
+    if s.is_empty() {
+        Ok(None)
+    } else if !s.starts_with("0x") {
+        Err(serde::de::Error::custom("expected hex string"))
+    } else {
+        u64::from_str_radix(&s[2..], 16).map(Some).map_err(serde::de::Error::custom)
+    }
+}

objdiff-core/src/bindings/wasm.rs

@@ -0,0 +1,53 @@
+use anyhow::Context;
+use prost::Message;
+use wasm_bindgen::prelude::*;
+
+use crate::{bindings::diff::DiffResult, diff, obj};
+
+#[wasm_bindgen]
+pub fn run_diff(
+    left: Option<Box<[u8]>>,
+    right: Option<Box<[u8]>>,
+    config: diff::DiffObjConfig,
+) -> Result<String, JsError> {
+    let target = left
+        .as_ref()
+        .map(|data| obj::read::parse(data, &config).context("Loading target"))
+        .transpose()
+        .map_err(|e| JsError::new(&e.to_string()))?;
+    let base = right
+        .as_ref()
+        .map(|data| obj::read::parse(data, &config).context("Loading base"))
+        .transpose()
+        .map_err(|e| JsError::new(&e.to_string()))?;
+    let result = diff::diff_objs(&config, target.as_ref(), base.as_ref(), None)
+        .map_err(|e| JsError::new(&e.to_string()))?;
+    let left = target.as_ref().and_then(|o| result.left.as_ref().map(|d| (o, d)));
+    let right = base.as_ref().and_then(|o| result.right.as_ref().map(|d| (o, d)));
+    let out = DiffResult::new(left, right);
+    serde_json::to_string(&out).map_err(|e| JsError::new(&e.to_string()))
+}
+
+#[wasm_bindgen]
+pub fn run_diff_proto(
+    left: Option<Box<[u8]>>,
+    right: Option<Box<[u8]>>,
+    config: diff::DiffObjConfig,
+) -> Result<Box<[u8]>, JsError> {
+    let target = left
+        .as_ref()
+        .map(|data| obj::read::parse(data, &config).context("Loading target"))
+        .transpose()
+        .map_err(|e| JsError::new(&e.to_string()))?;
+    let base = right
+        .as_ref()
+        .map(|data| obj::read::parse(data, &config).context("Loading base"))
+        .transpose()
+        .map_err(|e| JsError::new(&e.to_string()))?;
+    let result = diff::diff_objs(&config, target.as_ref(), base.as_ref(), None)
+        .map_err(|e| JsError::new(&e.to_string()))?;
+    let left = target.as_ref().and_then(|o| result.left.as_ref().map(|d| (o, d)));
+    let right = base.as_ref().and_then(|o| result.right.as_ref().map(|d| (o, d)));
+    let out = DiffResult::new(left, right);
+    Ok(out.encode_to_vec().into_boxed_slice())
+}

objdiff-core/src/diff/code.rs

@@ -1,8 +1,4 @@
-use std::{
-    cmp::max,
-    collections::BTreeMap,
-    time::{Duration, Instant},
-};
+use std::{cmp::max, collections::BTreeMap};
 
 use anyhow::{anyhow, Result};
 use similar::{capture_diff_slices_deadline, Algorithm};
@@ -100,13 +96,8 @@ fn diff_instructions(
     left_code: &ProcessCodeResult,
     right_code: &ProcessCodeResult,
 ) -> Result<()> {
-    let deadline = Instant::now() + Duration::from_secs(5);
-    let ops = capture_diff_slices_deadline(
-        Algorithm::Patience,
-        &left_code.ops,
-        &right_code.ops,
-        Some(deadline),
-    );
+    let ops =
+        capture_diff_slices_deadline(Algorithm::Patience, &left_code.ops, &right_code.ops, None);
     if ops.is_empty() {
         left_diff.extend(
             left_code

objdiff-core/src/diff/data.rs

@@ -1,7 +1,4 @@
-use std::{
-    cmp::{max, min, Ordering},
-    time::{Duration, Instant},
-};
+use std::cmp::{max, min, Ordering};
 
 use anyhow::{anyhow, Result};
 use similar::{capture_diff_slices_deadline, get_diff_ratio, Algorithm};
@@ -47,15 +44,13 @@ pub fn diff_data_section(
     left_section_diff: &ObjSectionDiff,
     right_section_diff: &ObjSectionDiff,
 ) -> Result<(ObjSectionDiff, ObjSectionDiff)> {
-    let deadline = Instant::now() + Duration::from_secs(5);
     let left_max =
         left.symbols.iter().map(|s| s.section_address + s.size).max().unwrap_or(0).min(left.size);
     let right_max =
         right.symbols.iter().map(|s| s.section_address + s.size).max().unwrap_or(0).min(right.size);
     let left_data = &left.data[..left_max as usize];
     let right_data = &right.data[..right_max as usize];
-    let ops =
-        capture_diff_slices_deadline(Algorithm::Patience, left_data, right_data, Some(deadline));
+    let ops = capture_diff_slices_deadline(Algorithm::Patience, left_data, right_data, None);
     let match_percent = get_diff_ratio(&ops, left_data.len(), right_data.len()) * 100.0;
 
     let mut left_diff = Vec::<ObjDataDiff>::new();
@@ -157,9 +152,7 @@ pub fn diff_data_symbol(
     let right_data = &right_section.data[right_symbol.section_address as usize
         ..(right_symbol.section_address + right_symbol.size) as usize];
 
-    let deadline = Instant::now() + Duration::from_secs(5);
-    let ops =
-        capture_diff_slices_deadline(Algorithm::Patience, left_data, right_data, Some(deadline));
+    let ops = capture_diff_slices_deadline(Algorithm::Patience, left_data, right_data, None);
     let match_percent = get_diff_ratio(&ops, left_data.len(), right_data.len()) * 100.0;
 
     Ok((
@@ -209,15 +202,9 @@ pub fn diff_bss_section(
     left_diff: &ObjSectionDiff,
     right_diff: &ObjSectionDiff,
 ) -> Result<(ObjSectionDiff, ObjSectionDiff)> {
-    let deadline = Instant::now() + Duration::from_secs(5);
     let left_sizes = left.symbols.iter().map(|s| (s.section_address, s.size)).collect::<Vec<_>>();
     let right_sizes = right.symbols.iter().map(|s| (s.section_address, s.size)).collect::<Vec<_>>();
-    let ops = capture_diff_slices_deadline(
-        Algorithm::Patience,
-        &left_sizes,
-        &right_sizes,
-        Some(deadline),
-    );
+    let ops = capture_diff_slices_deadline(Algorithm::Patience, &left_sizes, &right_sizes, None);
     let mut match_percent = get_diff_ratio(&ops, left_sizes.len(), right_sizes.len()) * 100.0;
 
     // Use the highest match percent between two options:

objdiff-core/src/diff/display.rs

@@ -12,7 +12,7 @@ pub enum DiffText<'a> {
     /// Colored text
     BasicColor(&'a str, usize),
     /// Line number
-    Line(usize),
+    Line(u32),
     /// Instruction address
     Address(u64),
     /// Instruction mnemonic
@@ -49,7 +49,7 @@ pub fn display_diff<E>(
         return Ok(());
     };
     if let Some(line) = ins.line {
-        cb(DiffText::Line(line as usize))?;
+        cb(DiffText::Line(line))?;
     }
     cb(DiffText::Address(ins.address - base_addr))?;
     if let Some(branch) = &ins_diff.branch_from {

objdiff-core/src/diff/mod.rs

@@ -1,6 +1,7 @@
 use std::collections::HashSet;
 
 use anyhow::Result;
+use wasm_bindgen::prelude::wasm_bindgen;
 
 use crate::{
     diff::{
@@ -17,6 +18,7 @@ pub mod code;
 pub mod data;
 pub mod display;
 
+#[wasm_bindgen]
 #[derive(
     Debug,
     Copy,
@@ -41,6 +43,7 @@ pub enum X86Formatter {
     Masm,
 }
 
+#[wasm_bindgen]
 #[derive(
     Debug,
     Copy,
@@ -65,6 +68,7 @@ pub enum MipsAbi {
     N64,
 }
 
+#[wasm_bindgen]
 #[derive(
     Debug,
     Copy,
@@ -93,6 +97,7 @@ pub enum MipsInstrCategory {
     R5900,
 }
 
+#[wasm_bindgen]
 #[derive(
     Debug,
     Copy,
@@ -117,6 +122,7 @@ pub enum ArmArchVersion {
     V6K,
 }
 
+#[wasm_bindgen]
 #[derive(
     Debug,
     Copy,
@@ -148,6 +154,7 @@ pub enum ArmR9Usage {
 #[inline]
 const fn default_true() -> bool { true }
 
+#[wasm_bindgen]
 #[derive(Debug, Clone, Eq, PartialEq, serde::Deserialize, serde::Serialize)]
 #[serde(default)]
 pub struct DiffObjConfig {
@@ -200,6 +207,9 @@ impl DiffObjConfig {
     }
 }
 
+#[wasm_bindgen]
+pub fn default_diff_obj_config() -> DiffObjConfig { Default::default() }
+
 #[derive(Debug, Clone)]
 pub struct ObjSectionDiff {
     pub symbols: Vec<ObjSymbolDiff>,

objdiff-core/src/lib.rs

@@ -1,4 +1,5 @@
 pub mod arch;
+pub mod bindings;
 #[cfg(feature = "config")]
 pub mod config;
 pub mod diff;

objdiff-core/src/obj/mod.rs

@@ -41,7 +41,7 @@ pub struct ObjSection {
     pub relocations: Vec<ObjReloc>,
     pub virtual_address: Option<u64>,
     /// Line number info (.line or .debug_line section)
-    pub line_info: BTreeMap<u64, u64>,
+    pub line_info: BTreeMap<u64, u32>,
 }
 
 #[derive(Debug, Clone, Eq, PartialEq)]
@@ -103,7 +103,7 @@ pub struct ObjIns {
     pub reloc: Option<ObjReloc>,
     pub branch_dest: Option<u64>,
     /// Line number
-    pub line: Option<u64>,
+    pub line: Option<u32>,
     /// Formatted instruction
     pub formatted: String,
     /// Original (unsimplified) instruction
@@ -136,8 +136,8 @@ pub struct ObjExtab {
 
 pub struct ObjInfo {
     pub arch: Box<dyn ObjArch>,
-    pub path: PathBuf,
-    pub timestamp: FileTime,
+    pub path: Option<PathBuf>,
+    pub timestamp: Option<FileTime>,
     pub sections: Vec<ObjSection>,
     /// Common BSS symbols
     pub common: Vec<ObjSymbol>,

objdiff-core/src/obj/read.rs

@@ -426,7 +426,7 @@ fn line_info(obj_file: &File<'_>, sections: &mut [ObjSection]) -> Result<()> {
             };
             let end = start + size as u64;
             while reader.position() < end {
-                let line_number = read_u32(obj_file, &mut reader)? as u64;
+                let line_number = read_u32(obj_file, &mut reader)?;
                 let statement_pos = read_u16(obj_file, &mut reader)?;
                 if statement_pos != 0xFFFF {
                     log::warn!("Unhandled statement pos {}", statement_pos);
@@ -468,7 +468,7 @@ fn line_info(obj_file: &File<'_>, sections: &mut [ObjSection]) -> Result<()> {
                 let mut rows = program.rows();
                 while let Some((_header, row)) = rows.next_row()? {
                     if let (Some(line), Some(lines)) = (row.line(), &mut lines) {
-                        lines.insert(row.address(), line.get());
+                        lines.insert(row.address(), line.get() as u32);
                     }
                     if row.end_sequence() {
                         // The next row is the start of a new sequence, which means we must
@@ -600,7 +600,14 @@ pub fn read(obj_path: &Path, config: &DiffObjConfig) -> Result<ObjInfo> {
         let timestamp = FileTime::from_last_modification_time(&file.metadata()?);
         (unsafe { memmap2::Mmap::map(&file) }?, timestamp)
     };
-    let obj_file = File::parse(&*data)?;
+    let mut obj = parse(&data, config)?;
+    obj.path = Some(obj_path.to_owned());
+    obj.timestamp = Some(timestamp);
+    Ok(obj)
+}
+
+pub fn parse(data: &[u8], config: &DiffObjConfig) -> Result<ObjInfo> {
+    let obj_file = File::parse(data)?;
     let arch = new_arch(&obj_file)?;
     let split_meta = split_meta(&obj_file)?;
     let mut sections = filter_sections(&obj_file, split_meta.as_ref())?;
@@ -616,7 +623,7 @@ pub fn read(obj_path: &Path, config: &DiffObjConfig) -> Result<ObjInfo> {
     line_info(&obj_file, &mut sections)?;
     let common = common_symbols(arch.as_ref(), &obj_file, split_meta.as_ref())?;
     let extab = exception_tables(&mut sections, &obj_file)?;
-    Ok(ObjInfo { arch, path: obj_path.to_owned(), timestamp, sections, common, extab, split_meta })
+    Ok(ObjInfo { arch, path: None, timestamp: None, sections, common, extab, split_meta })
 }
 
 pub fn has_function(obj_path: &Path, symbol_name: &str) -> Result<bool> {